Missing data in controlled clinical trials may create uncertainty in the results of a study based on non-missing data. We used 4 approaches of sensitivity analysis to address this problem. Radiographic progression data were used from a randomized controlled trial of patients with rheumatoid arthritis treated with leflunomide, methotrexate, or placebo for 12 months as an example. The mean change from baseline of the Sharp total radiographic score was the primary efficacy variable for the evaluation of leflunomide in comparison with placebo in the retardation of radiographic progression. Computer simulations were used in some of these approaches. The proportion of missing radiographic data was 26.4%. Result from the non-missing data showed that leflunomide was highly statistically significantly better than placebo in the retardation of radiographic progression. Results from the sensitivity analysis showed that radiographic data are sufficiently robust that it is unlikely that the missing data would have changed the conclusions from the analysis based on non-missing data. The potential effect of missing data in the results of a clinical trial may be addressed by various methods of sensitivity analysis. Computer simulation can be a useful tool in some of these approaches.